Skip to main content

Table 1 Performance of four metagenomic classifiers (BugSeq, MetaMaps, Centrifuge, and CDKAM) on GridION ZymoBIOMICS Mock Log and Even communities

From: BugSeq: a highly accurate cloud platform for long-read metagenomic analyses

 

Dataset used

F1 score at species level

Precision at species level (%)

Recall at species level (%)

Number of unique species detected (true count = 10)

Run time (wall clock) (hh:mm)

Classification memory requirement (GB)

Indexing time (wall clock) (hh:mm)

Indexing memory requirement (GB)

BugSeq

Even

0.95

99.82

90.36

117

4:09

118

1:15

243

 

Log

0.95

100

90.88

52

4:25

113

  

MetaMaps (miniSeq + H)

Even

0.89

99.13

80.71

2144

86:20

172.6

Prebuilt

Prebuilt

 

Log

0.94

99.46

88.25

1386

125:22

184.0

  

MetaMaps (RefSeq)

Even

Out of memory

NA

NA

     
 

Log

0.94

99.73

88.66

2094

154:29

270

  

Centrifuge – 16 bp minimum hit length, -k 1

Even

0.79

93.00

69.17

4337

00:19

67.7

19:45

168

 

Log

0.92

95.93

89.07

5777

00:19

64.4

  

Centrifuge – default (22 bp) minimum hit length, -k 1

Even

0.80

94.71

68.53

5380

00:16

56.9

  
 

Log

0.93

97.14

88.65

5513

00:14

55.7

  

CDKAM (standard + human)

Even

0.91

96.52

86.53

3721

00:09

34.5

28:23

66

 

Log

0.94

99.06

89.2

2956

00:08

34.5

  
  1. Different k-mer seed sizes were explored for Centrifuge (16 bp and 22 bp), and two different databases were examined for MetaMaps (miniSeq + H and RefSeq); otherwise, default parameters were used. The best result for each column and dataset combination is bolded